Highlights

Last Friday was slow news day in AI (rare!), so instead of scraping around for non-stories, Kyle compiled his complete learning roadmap. This is what he actually used to get from zero to teaching companies about AI at $2,000/hour. No fluff, just the resources that work.

If you work through these resources over a couple of weeks, you'll know more about large language models than 99% of people out there. For most people, it's magic. You'll realise it's all mathematics.

🎬 Start Here: Fun Videos That Actually Teach

Discussed at [00:01:28]

Casually Explained - "The Levels of AI"

Covers narrow vs general v super intelligence (most people still mix these up). Fun, silly but still useful.

Exurb1a - "How Will We Know When AI is Conscious?"

3Blue1Brown - LLMs explained

The mathematics without the pain. You're not going to be handwriting differential equations don’t worry! Start here if math scares you. These make complex concepts intuitive without requiring a PhD.

If you want to go deeper then 3Blue1Brown’s whole playlist on neural networks is fantastic.

🎓 “The Karpathy Trilogy”

Discussed at [00:07:26]

Andrej Karpathy's Essential Videos:

1. Intro to Large Language Models (1 hour) - Start here

2. Deep Dive into LLMs (3.5 hours) - The main course

3. How I Use LLMs (2+ hours) - Practical application

I personally went through all if this on the treadmill! It's long but chunk it up.

🧠 Foundational Books: Understanding What's Actually Happening

Discussed at [00:05:07]

Essential Reading:

"What Is ChatGPT Doing... and Why Does It Work?" by Stephen Wolfram

• Available free online (or get the book)

• Written when GPT-3.5 launched but still the best explainer

• Gets progressively complex but starts accessible

"How AI Works" by Ronald Kneusel

• Thin book, light language, deep insights. Available on Amazon

• Perfect bridge between pop-sci and technical

Pro tip: ChatGPT knows these texts well. Use it as your tutor: "I'm reading Wolfram's piece and don't understand [this part]"

🏫 Structured Learning: Free Courses

Discussed at [00:09:11]

Deep Learning AI (deeplearning.ai)

• Start with: "Generative AI for Everyone" (10 hours total)

• then there are multiple pathways you can take from there depending on your interests and what you want to do with AI.

• All free.

Platform Academies:

Both OpenAI and Anthropic have academies. I personally find OpenAI’s to be a bit of a mess and prefer Anthropic’s as it is less bloated and easier to find useful content.

📖 Going Deeper: The AI Canon

Discussed at [00:12:58]

For Those Ready to Dive Deep:

The AI Canon (a16z.com)

• All seminal AI papers in one place

• Includes Karpathy's "Software 2.0"

• Transformers papers, stable diffusion, etc.

• Comes with an Airtable tracker

Here’s an Airtable with all the resources in one place.

Some recommend this as a starting place. Nope. Not at all. It’s got some heavy stuff! Warm yourself up first with the previous resources.

📰 Staying Current: Signal vs Noise

Discussed at [00:18:09]

Kyle's Recommended Sources:

1. Nate B Jones (@natebjones)

   • TikTok

   • Substack

2. Ethan Mollick

   • Substack

   • Twitter

3. The Rundown AI

   • Daily AI news

Well be back tomorrow with our usual news updates! Hopefully this resources guide is a handy one off- make sure to bookmark and refer back to it as there’s a lot here!

Kyle

Why it matters: The skinny on what's happening in AI - straight from the previous live session: Highlights Last Friday was slow news day in AI (rare!), so instead of...